A project guide for using Intel GPUs in a Kubernetes cluster, including device plugins, video processing, inference, and GPU monitoring. It documents device names and access rules that workloads depend on.

Use it when changing Intel GPU device exposure, VA-API video processing, GPU inference workloads, or Grafana GPU metrics and alerts.

It prevents changes from making video transcoding fail silently or locking workloads out of the correct GPU devices. It also explains why video processing and AI inference may react differently to the same device change.

How Aviator-Coding/home-ops configures Claude Code and Codex

Reuse

Borrowing it

Nothing to install: this file belongs to Aviator-Coding/home-ops. Take a copy, put it at the same path in your own repository, and replace the rules that are about this project with yours.

Copy the file

curl -O https://raw.githubusercontent.com/Aviator-Coding/home-ops/main/.agents/skills/intel-gpu/SKILL.md

Clone the repo

git clone --depth 1 https://github.com/Aviator-Coding/home-ops

Made for: Claude Code, Codex.

Wrote this? Show the measurements

A badge with what this costs and how it scanned, read live from this page, so it follows the numbers instead of freezing them. Markdown for a README, HTML for a documentation site or a project page.

README.md

[![agentmods](https://agentmods.dev/badge/skills/aviator-coding/home-ops/intel-gpu/github.svg)](https://agentmods.dev/skills/aviator-coding/home-ops/intel-gpu)

Your own site

<a href="https://agentmods.dev/skills/aviator-coding/home-ops/intel-gpu"><img src="https://agentmods.dev/badge/skills/aviator-coding/home-ops/intel-gpu/github.svg" alt="Measured on agentmods" height="20"></a>

Or the 80×15 button, for a site that already has a row of RSS and ATOM ones. Only the verdict fits; the numbers stay here.

Your own site · 80×15

<a href="https://agentmods.dev/skills/aviator-coding/home-ops/intel-gpu"><img src="https://agentmods.dev/badge/skills/aviator-coding/home-ops/intel-gpu.svg" alt="Reviewed on agentmods" width="80" height="20"></a>

Per session 93 Skills are progressive disclosure: only the name and description are preloaded; the body loads when the skill is used.

When invoked 1,389 The whole file, excluding the scripts and references it only reads on demand.

Security scan A 0 findings. A grade says what 26 rules found in the file — not that it is safe.

Origin original No closer match found in the catalogue.

Token cost

What it costs to keep this loaded

Counted locally with the o200k_base tokenizer, which is exact for GPT models; Claude uses its own tokenizer and its counts differ. Treat this as one consistent yardstick across the catalogue rather than a bill. Prices are per million input tokens.

Model	Per session	Once invoked
Fable 5.1	$0.00093	$0.01389
Opus 5.5	$0.00037	$0.00556
Sonnet 5.5	$0.00019	$0.00278
Haiku 4.5	$0.00009	$0.00139

Measured 23d ago against content hash 122d94030dca, method: parsed. Prices are Anthropic first-party input rates as of 2026-10-07, from the pricing page.

Security

Grade A, and why

intel-gpu scanned grade A with 0 findings against 26 rules in 11 categories — prompt injection, anti-refusal, data exfiltration, privilege escalation, supply chain, agent snooping, system-prompt leakage, SSRF and excessive agency — measured 23d ago.

A static scan of the body, not an audit. Every finding is printed with the line that produced it so you can judge whether it matters here. A mod is markdown that instructs an agent; that is exactly why what it instructs is worth reading.

Nothing flagged

None of the 26 patterns this scan looks for appear in this file: no shell pipes, no recursive deletes, no credential paths, no hidden text, no instruction-override or anti-refusal phrasing, no agent-config snooping. That is not a guarantee, it is the absence of the things that are checkable.

.agents/skills/intel-gpu/SKILL.md · 85 lines

How it starts

The opening of the file, as written. The whole thing — 85 lines — stays where its author put it; the contents beside it link to each section on GitHub.

Intel GPUs: devices, VA-API, telemetry

One discrete Arc Pro B70 on talos-3 (PCI 0000:03:00.0, 8086:e223) plus a Raptor Lake iGPU (0xa7a0) on each node. llama.cpp flags, the NaN probe, and the embedding rate curve are skill b70-llm-serving. This skill is the device and the metrics.

Tripwires

Never rename a DRM node with generic-device-plugin mountPath. The b70 group exposes the card as card0 / renderD128 for Level Zero. libdrm ignores that path, fstats the fd, and reopens the kernel DEVNAME (dri/renderD129). VA-API then fails before a driver loads. Level Zero stays green, so inference looks healthy while transcoding is dead. VA-API consumers use devic.es/b70-vaapi (card1 / renderD129). Allocatable capacity is not proof. devices.md
Device IDs are sha1(count + every host path in the group). Adding a path to b70 changes all 99 IDs and drops kubelet's live allocations. Add a group instead of editing that one.
The device-plugin config is a subPath mount. Kubelet never refreshes it. The configMapGenerator name-suffix hash in kubernetes/apps/base/system/generic-device-plugin/app/kustomization.yaml must stay enabled. A comment-only edit of config/config.yaml still rolls the DaemonSet. Do not touch that file to tidy prose.
No per-engine utilization exporter. Prometheus has the B70 xe hwmon chip (temps, fan, energy counters) and kube-state-metrics allocation for the iGPU. The iGPU has no hwmon chip. There is no VRAM-utilization gauge and no xpu-smi equivalent. telemetry.md
Do not migrate GPU scheduling to DRA, and never via adminAccess: true. The Intel driver README says it is beta and not for production, and it cannot share one GPU (upstream issue #79, strict 1-to-1 or SR-IOV). vllm and tdarr-node both need the one B70. Reopen only when the CAUTION line is gone and #79 ships allowMultipleAllocation: true. dra.md
allowIDs: "0xa7a0" is already set on the Intel GPU plugin (kubernetes/apps/base/system/intel-device-plugin-operator/gpu/helmrelease.yaml). It scopes gpu.intel.com/xe to the iGPU. Do not re-add it, and do not drop it while the B70 is also an xe device.
Do not buy a second B70 on the current recommendation. The buy-if is a >32 GB MoE in production and talos-3 seating that card at PCIe x8 or wider, or llm-scaler#382 closing so vLLM-XPU prefill is a hard requirement. The chassis question (second OCuLink or x8/x8 bifurcation) is still open. second-card.md

Read the full file on GitHub · 85 lines

Files

What ships with it

6 files beside SKILL.md in the same directory: the scripts, references and assets a skill reads on demand. Not counted in the per-session cost; read them before you install if any of them is executable.

Changes

What this file has done since we first saw it

Hashed on every crawl. A supply-chain change to an agent config is a question of when, not whether, so the history is kept rather than the latest state alone.

23d ago Changed · +1 lines 122d94030dca
25d ago Changed · +62 lines a153a07ac9d9
26d ago Changed · +6 lines c0927d2a6cbd
1mo ago Changed e9bd784a0324
1mo ago First seen · 16 lines · 93 tokens per session scan A f648e0ffc325

Subscribe to this mod's changes

intel-gpu is a skill published in the GitHub repository Aviator-Coding/home-ops (3 stars, last pushed today), licensed MIT. It adds 93 tokens to every session and 1,389 once invoked, about $0.0004 per session on Opus 5.5. A static security scan graded it A with 0 findings. No closer match exists in the catalogue, so it is treated as the original; first seen 2026-09-04.

DevOps and cloud Hardware and IoT

gke-compute-classes

Configures, optimizes, and troubleshoots GKE ComputeClasses. Use when configuring Spot VMs with on-demand fallback, targeting specific accelerators (GPUs/TPUs) or machine families, restricting ComputeClass access, or debugging pending pods related to node pool auto-creation. Do not use for cluster-level Node Auto…

google/skills · 83 tokens

jetson-diagnostic

Read-only Jetson health snapshot for identity, memory, GPU, thermal, power, storage, services, and top processes.

NVIDIA/skills · 30 tokens

holoscan-install-wheel

Install Holoscan SDK 5.x CUDA 13 Python wheels in an isolated environment, choosing the full stack or version-matched components.

NVIDIA/skills · 34 tokens

offensive-z-wave

Z-Wave attack methodology — sniffing with Z-Force / EZ-Wave / RTL-SDR + ZniffMobile, S0 (legacy) network-key derivation flaw and key reuse, S2 (modern) ECDH commissioning analysis, replay/injection on unauthenticated nodes, default-key brute-force on test deployments, and home-automation hub pivots. Use when targeting…

SnailSploit/Claude-Red · 113 tokens

doca-socket-relay

Use this skill when the operator is driving the DOCA Socket Relay to bridge a socket-oriented host application onto a BlueField DPU peer without rewriting it — picking the deployment shape (in-process, sidecar, or BlueField service container), configuring the host-side socket and the DPU-side forwarding endpoint…

NVIDIA/skills · 236 tokens

hsb-flash

Flash the FPGA on an HSB board connected to an NVIDIA devkit. Supports HSB Lattice boards (FPGA versions 2407, 2412, 2507, 2510) and Leopard Imaging VB1940 "all-in-one" cameras (FPGA versions 2507, 2510). Uses release-specific YAML manifests and board-type-specific program commands. Lattice and VB1940 commands must…

NVIDIA/skills · 94 tokens

home-ops: Skill for Claude Code